[ 1 ]
Word Completion and Sequence Prediction in Bangla Language Using Trie and a Hybrid Approach of Sequential LSTM and N-gram
IEEE · ICAICT 2020
2nd International Conference on Advanced Information and Communication
Technology (ICAICT)
(Published: Feb 2021)
Bangla NLP
Sequence Prediction
Trie Architecture
Sequential LSTM
N-gram Models
Encoder-Decoder
Text composition efficiency in low-resource and morphologically rich scripts such as Bangla often suffers from cumbersome keyboard navigation. This paper introduces a hybrid sequence prediction and word completion pipeline fusing hierarchical Trie search structures with sequential Long Short-Term Memory (LSTM) recurrent networks and N-gram statistical models. Evaluated across large-scale Bangla text corpora, the hybrid system demonstrates significant improvements in keystroke savings and prediction accuracy compared to standalone statistical or neural baselines.
@inproceedings{sarker2020word,
author = {Sarker, Soumik and Islam, Md. Ekramul and Saurav, Jillur Rahman and Nahid, Md Mahadi Hasan},
title = {Word Completion and Sequence Prediction in Bangla Language Using Trie and a Hybrid Approach of Sequential LSTM and N-gram},
booktitle = {2020 2nd International Conference on Advanced Information and Communication Technology (ICAICT)},
year = {2020},
pages = {174--179},
publisher = {IEEE},
doi = {10.1109/ICAICT51780.2020.9333518}
}